REPORT EMBARGOED UNTIL 2:00 PM US EASTERN TIME THURSDAY, 26 FEBRUARY 2009 trail of two prints and a single isolated hominin print (Fig. 3). The footprints occur within a 9-m- thick sequence of fine-grained, normally graded, Early Hominin Foot Morphology Based silt and sand units deposited as overbank flood deposits with evidence of paleosol development. Interbedded within this succession are three flu- on 1.5-Million-Year-Old Footprints vially reworked volcanic ashes; the upper ash (Northern Ileret Tuff) forms a prominent land- from Ileret, scape bench that correlates with other nearby sites where traces of hominin activity have been 1 2 3,4 5 6 Matthew R. Bennett, * John W.K. Harris, Brian G. Richmond, David R. Braun, Emma Mbua, recovered (15) and is unconformably overlain 6 7 6 7 Purity Kiura, Daniel Olago, Mzalendo Kibunjia, Christine Omuombo, by the Galana Boi Formation of Holocene age 8 9 9 Anna K. Behrensmeyer, David Huddart, Silvia Gonzalez (12). The ash layers are correlated geochemi- cally to dated tuffs within the Turkana Basin, Hominin footprints offer evidence about gait and foot shape, but their scarcity, combined with an thereby providing an age of 1.51 to 1.52 Ma for inadequate hominin fossil record, hampers research on the evolution of the human gait. Here we the upper tuff and 1.53 Ma for the lower tuff report hominin footprints in two sedimentary layers dated at 1.51 to 1.53 million years ago (Ma) at (Fig. 1) (14, 16). Ileret, Kenya, providing the oldest evidence of an essentially modern human–like foot anatomy, The prints from both the upper and lower with a relatively adducted hallux, medial longitudinal arch, and medial weight transfer before levels at FwJj14E have a well-defined, deeply push-off. The size of the Ileret footprints is consistent with stature and body mass estimates for depressed and adducted hallux; visible lateral toe /erectus, and these prints are also morphologically distinct from the 3.75-million- impressions that vary in depth of indentation; a year-old footprints at , Tanzania. The Ileret prints show that by 1.5 Ma, hominins had well-defined ball beneath the first and second evolved an essentially modern human foot function and style of bipedal locomotion. metatarsal heads; and a visible instep reflecting a medial longitudinal arch. The angle of hallux ipedalism is a key human adaptation that heads, and finally ending with toe-off pressure abduction, relative to the long axis of the foot, is appears in the fossil record by 6 million under the hallux (9, 10). As a consequence, the typically 14° compared to, and statistically dis- Byears ago (Ma) (1). Considerable debate deepest part of a footprint often occurs beneath tinct from (table S4), 8° for the modern reference continues over when and in what context a mod- the first and second metatarsal heads, that along prints and 27° for the Laetoli prints (Fig. 4A). ern human–like form of evolved, be- with a deep hallucal impression corresponds to The morphology of the Ileret prints suggests that cause of a fragmentary record and disagreements the peak pressures at toe-off (10). The extent to the feet of these hominins had functional medial over the functional interpretations of existing which any pressure, or footprint impression, oc- longitudinal arches. In prints FUT1-2 and FLI1, fossils and footprints (2–7). Modern human foot- curs medially varies with the anatomy of the mid- for example, the medial side of the mid-print is prints reflect the specialized anatomy and func- foot, including the height of the longitudinal arch slightly raised, indicating a lack of substrate im- tion of the human foot, which is characterized by and other factors (11), and the extent to which pression in this area (Figs. 2 and 3). A com- a fully adducted hallux, a large and robust cal- lateral toes leave impressions depends on factors parison of the instep width relative to the width in caneus and tarsal region, a pronounced medial such as foot orientation relative to the direction of the metatarsal head region shows that the upper longitudinal arch, and short toes (2). Footprints travel, transverse versus oblique push-off axes, prints at FwJj14E fall within the modern human reflect the pressure distribution as the foot makes and substrate properties. This contrasts with the range and are distinct from the relatively wider contact with the substrate, but also the sediment’s less stereotypical pattern of footfall observed in insteps characterizing the Laetoli prints (Fig. 4C). geomechanical properties (8). During normal African apes during quadrupedal and bipedal The FwJj14E lower prints show more variability; walking, the weight-bearing foot undergoes a locomotion. Here the heel and lateral mid-foot the two prints that differ from modern human highly stereotypical movement and pressure dis- make contact with the ground first, followed by prints appear to have undergone taphonomic tribution pattern in which the heel contacts the contact with the lateral toes that are often curled mediolateral compression (fig. S13). Most of the ground first, making a relatively deep impression and with a hallux that is often widely abducted. Ileret prints are similar in length to the longest on the substrate. This is followed by contact with Lift-off in the African apes is variable, but it modern human prints (fig. S19), although the the lateral side of the foot and metatarsal heads, usually involves relatively low pressure during isolated footprint on the lower level (FLI1; Fig. 3) after which weight transfers to the ball of the foot final contact by both the lateral toes and widely is significantly smaller, despite a similar gross with peak pressure under the medial metatarsal abducted hallux, in stark contrast to modern hu- anatomy, and may represent a subadult. man foot function (11). The contours of the upper- and lower-level 1School of Conservation Sciences, Bournemouth University, Here we report hominin footprints from FwJj14E prints suggest a modern human–like 2 Poole, BH12 5BB, UK. Department of Anthropology, Rutgers the Okote Member of the Forma- toe-off mechanism, in contrast to the more am- University, 131 George Street, New Brunswick, NJ 08901, USA. 3Center for the Advanced Study of Hominid Paleobiology, tion (12), second in age only to the mid- biguous evidence from the Laetoli footprints. Department of Anthropology, George Washington University, Pliocene (3.7 Ma) Laetoli prints (13), located Both the upper and lower Ileret prints show the Washington, DC 20052, USA. 4Human Origins Program, Na- close to Ileret, Kenya (Fig. 1; site FwJj14E; lat- greatest depth in the metatarsal head region to be tional Museum of Natural History, Smithsonian Institution, ′ ′ ′′ ′ ′ ′′ 5 itude 4°18 18 44 N, longitude 36°16 16 16 E). typically medially located, in contrast to its lateral Washington, DC 20013-7012, USA. Department of Archae- The footprints are found in association with position in many of the Laetoli prints (fig. S19). ology, University of Cape Town, Private Bag, Rondebosch 7701, South Africa. 6National Museums of Kenya, Post Office Box animal prints on two stratigraphically separated In the majority of the modern human prints, the 40658-00100, Nairobi, Kenya. 7Department of Geology, Uni- levels and were digitized with an optical laser normal shift in pressure from lateral to medial in versity of Nairobi, Post Office Box 30197, Nairobi, Kenya. scanner (Fig. 1) (14). The upper surface contains late stance phase, with peak medial pressures, is 8 Department of Paleobiology, MRC 121, National Museum of three hominin footprint trails comprising two registered in the greater medial depth. However, Natural History, Smithsonian Institution, Washington DC, 20013-7012, USA. 9School of Biological and Earth Sciences, trails of two prints and one of seven prints, as the variability shows that fully modern feet dur- Liverpool John Moores University, Liverpool, L3 3AF, UK well as a number of isolated prints (Figs. 2 and ing walking can also produce prints with greater *To whom correspondence should be addressed. E-mail: 3 and figs. S3 and S6 to S11). The lower sur- lateral depth (fig. S19) (3, 6), demonstrating [email protected] face, approximately 5 m below, preserves one variability in the foot function, something that is

www.sciencemag.org SCIENCE VOL 000 MONTH 2009 1 MS no: RE1168132/CT/PALEO REPORT

further complicated by the geomechanicalEMBARGOED prop- UNTIL 2:00Inferences PM US about EASTERN stride areTIME possible THURSDAY, from the 26 FEBRUARYof approximately 2009 0.63 m/s (14). This is a slow erties of the substrate. Therefore, a laterally con- short trails at FwJj14E. The longest trail (FUT1) speed consistent with someone beginning to centrated depth cannot, in itself, rule out (for consists of seven prints, initially with three close- walk from a standing (or slowed) position and instance, in the Laetoli prints) a foot structure ly spaced prints suggesting an individual either the variability in step lengths may attest to the capable of medial weight transfer in late stance standing with feet astride or who slowed before challenges of walking on an uneven muddy sur- phase. However, the modern human–like relative walking forward with an increasing step length face already marred by a range of animal prints. instep width (Fig. 4C) and medially concentrated (Fig. 2). Typical stride lengths vary from 460 to Although precise foot-to-stature ratios are un- ball impression in the Ileret prints provide com- 785 mm, with step lengths in the range of 431 to known for early Pleistocene hominins, we use as pelling positive evidence of a medial longitudinal 536 mm in the direction of travel; foot angles a rough estimate ones developed for Australian arch and the medial pressure shift and push-off vary, but are typically parallel to the direction of Aborigines (18) and Kenyan Dassenach (14), on from the ball of the foot beneath the medial travel, diverging by 1° to 24°. Using the average the basis that their statures are adapted for a metatarsal heads. These are hallmarks of modern of the last three strides (xxxxxxxxx = 876 mm) semiarid environment. Using this relationship, human walking that can be related to a long stride and a hip height of 860 mm estimated from foot we estimate the average height of the individuals with extended lower limbs, key to the energetic length (258 mm, the average of the four clearest from the prints on the upper surface to be 1.75 T efficiency of human walking (17). prints of the FUT1 trail), we estimate a velocity 0.26 m and 1.76 T 0.26 m for those on the lower

Fig. 1. Location,stratigraphy,andfootprintsurfacesatIleret(FwJj14E).

2 MONTH 2009 VOL 000 SCIENCE www.sciencemag.org MS no: RE1168132/CT/PALEO REPORT surface, excluding the potentialEMBARGOED subadult (print UNTILFwJj14E 2:00 PM were US EASTERN made by Homo TIME ergaster/erectusTHURSDAY, 26 FEBRUARYuncovered four 2009 of the original seven prints and a FUI1) which gives a height of 0.92 T 0.13 m individuals. new print, and the two best-preserved examples (table S2). The large stature and mass estimates Behrensmeyer and Laporte (22) reported are comparable to those at FwJj14E (14). derived from the Ileret prints compare well with hominin footprints at site GaJi10 (latitude To further evaluate the morphology of the those of Homo ergaster/erectus on the basis of 3°44′44′15′′N, longitude 36°55′55′48′′E), 45 km prints at FwJj14E and compare them objectively postcranial remains and are significantly larger to the south of FwJj14E, in 1981. The footprint with samples of modern human and Laetoli than postcrania-based stature and mass estimates surface occurs below a prominent tuff, sampled footprints, we digitized 13 landmarks on each for Paranthropus boisei and (table and correlated here to the Akait Tuff, dated to footprint scan and used generalized Procrustes S3) (19–21), suggesting that the prints at 1.435 Ma (14). Re-excavation of these prints analysis to compare their shapes (14). Figure 4B

Fig. 2. Tessellated swath of optical laser scans of the main footprint trail on the upper footprint surface at FwJj14E. Color is rendered with 5-mm isopleths.

Fig. 3. Optical laser scan images color-rendered with 5-mm isopleths for footprints at both FwJj14E and GaJi10. (A)Isolated left foot (FUI1) on the upper footprint surface at FwJj14E. (B)Photograph of FUI8 on the upper foot- print surface at FwJj14E, showing good definition of the toe pads; the sec- ond toe is partially ob- scured by the third toe. (C) Second trail on the up- per footprint surface at FwJj14E, showing two left feet. (D) Third trail on the upper footprint surface at FwJj14E, showing a right and a left foot. (E)Print R3 from GaJi10 (22), re- excavatedandscannedas part of this investigation. (F) Partial print (FUT1-2) on the upper footprint surface at FwJj14E; the heel area has been re- moved by a later bovid print. (G) Print FLI1 on the lower footprint surface at FwJj14E, rendered with Note the locations of the pads of the small toes and the presence of a well- 55-mm alternating black and white isopleths. (H) Inverted image of the toe defined ball beneath the hallucial metatarsophalangeal joint. D1, xxxxx; area of print FUT1-1 with alternating 55-mm black and white isopleths. D3, xxxxx; D5, xxxxx.

www.sciencemag.org SCIENCE VOL 000 MONTH 2009 3 MS no: RE1168132/CT/PALEO REPORT

shows thin plate splines andEMBARGOED landmark vector UNTILboth 2:00 the PM modern US EASTERN and Ileret TIME prints, THURSDAY, possibly re- 26 FEBRUARYevident in some 2009Homo ergaster/erectus individu- maps comparing the mean landmark positions (in flecting differences in foot proportions or a lack als (19, 20). These prints add to the anatomical two dimensions) of the Ileret prints with the of definition of the instep. Discriminant analysis (19, 20, 23) and archaeological (24, 25) evidence modern human and Laetoli prints (14). When the was also used to compare the different print pointing to a major transition in human evolution prints from the two levels at FwJj14E are populations (14); 8 of the 10 prints at FwJj14E with the appearance of hominins with long lower compared, both surfaces show similar anatomical used in the analysis were classified with modern limbs, conferring advantages at a lower energetic differences from the modern prints, with a nar- prints and two with Laetoli prints (table S6). cost (26), and archaeological indications of ac- rower heel and ball area and a wider instep The Ileret footprints show the earliest evi- tivities in a variety of ecological settings and the associated with less pronounced arch elevation dence of a relatively modern human–like foot transport of resources over long distances (27). (Fig. 4B and fig. S20). When compared to the with an adducted hallux, a medial longitudinal These lines of evidence, together with the earliest Laetoli prints, the Ileret prints have a more con- arch, and medial weight transfer before push-off. evidence of a relatively modern foot anatomy and tracted proximal mid-foot region, including a Although we cannot conclude with certainty what function, support the hypothesis that this was a deeper instep (Fig. 4B), suggesting the presence hominin species made the footprints at FwJj14E hominin with a larger home range related to in- of a medial longitudinal arch. The location of the or GaJi10, these modern human characteristics, in creasing average body size and enhanced dietary narrowest point of the instep also lies farther combination with the large size of the prints, are quality (28). These factors add to an emerging forward (more distal) in the Laetoli prints than is most consistent with the large size and tall stature picture of the paleobiology of H. ergaster/erectus

A C

B

D

Fig. 4. (A) Angle of hallux abduction (14).Theblueboxshowsthe25thand75th percentiles (the median), whereas the outer lines mark the maximum and minimum values. (B) Contour maps (interval, 2 mm) for selected prints. Holocene H. sapiens prints are from the Sefton Coast in England (14). (C) Scatter plot of instep-to-ball width, showing the broader instep width of the Laetoli prints, see (A) for symbol key. (D) Thin plate splines and vector maps comparing the landmark anatomy for mean print forms. The vectors represent the direction of landmark movement between the two print populations. H, hxxxxx; I, ixxxxx; Idth, xxxxx; B, bxxxxxx; D, xxxxx; Hdth, hxxxx.

4 MONTH 2009 VOL 000 SCIENCE www.sciencemag.org MS no: RE1168132/CT/PALEO REPORT that suggests a shift in culturalEMBARGOED and biological UNTIL11. 2:00 E. Morag, PM US P. R.EASTERN Cavanagh, J. TIME Biomech. THURSDAY,32, 359 26 FEBRUARY25.B.L.Pobiner,M.J.Rogers,C.M.Monahan,J.W.K.Harris, 2009 adaptations relative to earlier hominins. (1999). J. Hum. Evol. 55, 103 (2008). 12. C. S. Feibel, F. H. Brown, I. McDougall, Am. J. Phys. 26. H. Pontzer, J. Exp. Biol. 210, 1752 (2007). Anthropol. 78, 595 (1989). 27. D. R. Braun, J. W. K. Harris, D. N. Maina, Archaeometry References and Notes 13. M. D. Leakey, R. L. Hay, Nature 278, 317 (1979). 10.1111/j.1475-4754.2008.00399 (2008). 1. B. G. Richmond, W. L. Jungers, Science 319, 1662 14. Materials and methods, additional illustrations, statistical 28. S. C. Antón, W. R. Leonard, M. L. Robertson, J. Hum. Evol. (2008). analysis, and discussion of tuff geochemistry at FwJj14E 43, 773 (2002). 2. N. L. Griffin, B. Wood, in The Human Foot: A Companion are available as supporting material on Science Online. 29. This project is part of the Koobi Fora Research and to Clinical Studies, L. Klenerman, B. Wood, Eds. 15. J. W. K. Harris, D. G. Braun, J. T. McCoy, B. L. Pobiner, Training Program, which is jointly run by the National (Springer-Verlag, London, 2006), pp. 27–79. M. J. Rogers, J. Hum. Evol. 42, A15 (2002). Museum of Kenya and Rutgers University and supported 3. T. D. White, G. Suwa, Am. J. Phys. Anthropol. 72, 485 16. F. H. Brown, B. Haileab, I. McDougall, J. Geol. Soc. by the Holt Family Foundation, M. Weiss, B. Mortenson, (1987). London 163, 185 (2006). G. Noland, C. Chip, and the Wenner-Gren Foundation for 4. R. H. Tuttle, D. Webb, E. Weidle, M. Baksh, J. Arch. Sci. 17. M. D. Sockol, D. A. Raichlen, H. Pontzer, Proc. Natl. Acad. Anthropological Research. B. Wood provided valuable 17, 347 (1990). Sci. U.S.A. 104, 12265 (2007). comments on a draft of the paper, and the dating was 5. J. Charteris, J. J. Wall, J. W. Nottrodt, Am. J. Phys. 18. S. Webb, M. L. Cupper, R. Robins, J. Hum. Evol. 50, 405 undertaken by F. Brown. The field assistance of Anthropol. 58, 133 (1982). (2006). H. Manley, S. A. Morse, and J. Steele is acknowledged. 6. D. J. Meldrum, in From Biped to Strider: The Emergence 19. C. B. Ruff, A. Walker, in The Nariokotome of Modern Human Walking, Running, and Resource Skeleton, A. C. Walker, R. E. F. Leakey, Eds. (Harvard Supporting Online Material Transport, D. J. Meldrum, C. E. Hilton, Eds. (Kluwer Univ. Press, Cambridge, MA, 1993), pp. 234–265. www.sciencemag.org/cgi/content/full/[vol]/[issue no.]/[page]/ Academic/Plenum Publishers, New York, 2004), 20. H. M. McHenry, K. Coffing, Annu. Rev. Anthropol. 29, DC1 pp. 63–83. 125 (2000). Methods 7. J. T. Stern Jr., R. L. Susman, Am. J. Phys. Anthropol. 60, 21. F. Spoor et al., Nature 448, 688 (2007). SOM Text 279 (1983). 22. A. K. Behrensmeyer, L. F. Laporte, Nature 289, 167 Figs. S1 to S22 8. J. R. L. Allen, Philos. Trans. R. Soc. London Ser. B 352, (1981). Tables S1 to S6 481 (1997). 23. B. G. Richmond, L. C. Aiello, B. A. Wood, J. Hum. Evol. References and Notes 9. H. Elftman, J. Manter, Am.J.Phys.Anthropol.20,69(1935). 43, 529 (2002). 10. E. Vereecke, K. D’Aout, D. De Clercq, L. Van Elsacker, 24. M. J. Rogers, J. W. K. Harris, C. S. Feibel, J. Hum. Evol. 5 November 2008; accepted 19 January 2009 P. Aerts, Am. J. Phys. Anthropol. 120, 373 (2003). 27, 139 (1994). 10.1126/science.1168132

www.sciencemag.org SCIENCE VOL 000 MONTH 2009 5 MS no: RE1168132/CT/PALEO